codegemma 7 B Manim Gen

Introduction

The CODEGEMMA-7B-MANIMGEN model, developed by Thanhkt, is a fine-tuned text generation model based on the unsloth/codegemma-7b-it-bnb-4bit model. It leverages the capabilities of the Hugging Face Transformers library and is optimized for faster training using the Unsloth library.

Architecture

The model architecture is built on top of Hugging Face's Transformers library, utilizing 4-bit precision for efficient computation. This setup enables the model to handle text generation tasks with lower resource consumption, maintaining performance while reducing computational overhead.

Training

The CODEGEMMA-7B-MANIMGEN model was trained at twice the usual speed by employing the Unsloth library and Hugging Face's TRL (Transformers Reinforcement Learning) library. This combination enhances the training process, providing a more efficient approach to model refinement.

Guide: Running Locally

1. Clone the Repository: Clone the model repository from Hugging Face to your local environment.
2. Install Dependencies: Ensure Python and essential libraries like transformers, torch, and safetensors are installed.
3. Load the Model: Use the Hugging Face Transformers library to load the model and tokenizer.

   from transformers import AutoModelForCausalLM, AutoTokenizer
   
   tokenizer = AutoTokenizer.from_pretrained("thanhkt/codegemma-7B-ManimGen")
   model = AutoModelForCausalLM.from_pretrained("thanhkt/codegemma-7B-ManimGen")
   

4. Run Inference: Input a text prompt and generate text using the model.

   inputs = tokenizer("Your text here", return_tensors="pt")
   outputs = model.generate(**inputs)
   print(tokenizer.decode(outputs[0]))
   

Cloud GPUs

For optimal performance, consider running the model on a cloud-based GPU service such as AWS, Google Cloud Platform, or Azure. This setup can handle the computational demands of model inference efficiently.

License

The CODEGEMMA-7B-MANIMGEN model is licensed under the Apache-2.0 License, which allows for both academic and commercial use, modification, and distribution.